Thinking about myself: xing and linked have largly replaced my local contacts database and my calendar is online, too; the acm-digital library, springer link, computer.org and google-scholar make me through away papers after I have read them (at the beginning of my PhD-studies I still sorted them in folders); gmail and gmx hold my private mail in the network; I have not bought paper maps for quite some time, and as recently posted access to knowledge is nowadays often via google. The current move of putting applications online – which I really like greatly – speeds up these trends.
With current sizes of hard drive and future network connectivity I think caching and recording becomes key. For many domains this is easy. Everything you have ever seen on the screen will be forever on the computer (easy for static content such as web pages, even for videos this is not far in the future; assuming 365 days x 8 hours x 1Gbyte/hour is about 3TByte/year). In many domains Pre-fetching seems useful. In some areas this seems straightforward. When you view a paper all papers that are cited and papers that cite this paper will be cached locally (and not just short term, but forever ;-), too.
I wonder when we can by the entire index of the web (e.g. google cache) for offline use. Will this every be possible or is content growing faster than storage?
At least I will start caching important information (e.g. mail & contacts) in the future locally.
From a business perspective this is interesting, too. Even if there is a major provider (e.g. Skype) people will create their own redundancies with a further provider (e.g. messenger) - so there will be always room for several players.